Correspondence to Suthasinee Kumluang; [email protected]
Strengths and limitations of this study
This study assessed the mean annual costs and all-cause mortality of patients with stroke with the most up-to-date nationally representative stroke cohort.
The two-part models and parametric survival analysis were performed.
The analysis was based on hospitalised stroke events; however, a small proportion of patients with less severe symptoms who were not admitted to hospital could have been missed in this study.
By using nationwide claim data, we included neither loss of productivity costs of patients nor unpaid care costs by their caregivers.
Introduction
Stroke represents the second leading cause of death in Thailand. Stroke mortality rates have been reported to be 38.7 to 47.1 (per 100 000) between 2014 and 20181 and subsequently accounted for 27 361 deaths in men and 23 669 in women in 2019.2 3 Although efforts to reduce the burden of stroke have been made, challenges of implementing stroke services still remain. Despite an increasing trend in thrombolysis provision, the absolute rate remains low,4 and is further exacerbated by a shortage of neurology specialists and a low referral rate to rehabilitation services.5 6
The Thai Ministry of Public Health (MOPH) has established the emergency medical service in 2008 and introduced the stroke fast track system for stroke care.7 The ‘ship-and-drip’ or ‘drip-and-ship’ and mothership models can be performed in Thailand8 9 but depend on hospital capacity and context. Thrombolysis treatment can be prescribed by a well-trained general practitioner under the supervision of a neurologist at a hub. Following this, patients are transferred to the hub for further treatment or moved to stroke unit (SU) - a comprehensive specialised service with multidisciplinary team and care specifically tailored to stroke patients - or stroke corner (SC) - a specialised area in an intensive unit or general wards - in the same hospital, if a spoke hospital has adequate capacity. In addition, the proportion of SUs in advanced-level and standard-level hospitals was 97% and 65%, respectively.10 Further details on stroke care are provided in online supplemental table 1.
In 2016, the Thai MOPH11 published their service plan strategy and a set of national stroke key performance indicators (KPIs) for 2018–2022, to increase service capacities and improve health outcomes, according to hospital levels in all 12 health regions across Thailand. Following which, in 2018, a report10 from the MOPH revealed that the quality of stroke care has improved and the number of hospitals providing a dedicated stroke unit has increased by 18% and 15% at advanced-level and standard-level hospitals, respectively. In addition, some mid-level hospitals can set-up SU/SC if hospitals have capacity and some community hospitals can set-up a rehabilitation ward. National stroke mortality rates have gradually declined from 8.2% in 2018 to 7.9% in 2020.10 12 However, long-term outcomes, such as health-related quality of life (HRQoL) post-stroke or long-term survival, are rarely monitored. Although it is well known that most stroke survivors are being discharged from hospital with disability as a consequence of their stroke,13 there is a lack of national-level information on service utilisation and health outcomes, so that a full assessment of the service plan strategy is difficult. This study aims to provide this national-level assessment by (1) estimating the resource utilisation of patients with stroke across stroke subtypes, and (2) estimating all-cause mortality of patients with incident stroke across stroke subtypes in Thailand.
Methods
Design
The national stroke data set (January 2017 to November 2020) was obtained from the National Health Security Office which is a health insurance organisation managing the universal coverage scheme (UCS) covering 75% of the Thai population. This data contained both outpatient and inpatient data, and covered contracted public and private hospitals throughout Thailand. Details on hospital level are provided in supplementary materials (online supplemental table 2).
All patients aged 18 years and over with either a principal diagnosis or secondary diagnosis of stroke were included in the study cohort. The cohort was identified using the International Classification of Diseases, 10th revision using code I60–I62 for haemorrhagic, I63 for ischaemic and I64–I69 for unspecified stroke. A 2-year lookback period (2015–2016) was used to identify incident strokes and to avoid double-counting of incident stroke events. All patients were identified from the first recorded hospital episode of stroke diagnosis and followed-up until death or end of their records. The data recorded for each hospital record include patient demographics, medical treatment information, hospital charges, out-of-pocket payments by patients and hospital reimbursement with adjusted relative weight per admission.
Cost estimation
An average cost for outpatient visit and inpatient day were obtained from a recent cost study in Thailand.14 For cost per inpatient admission, the unit cost from the cost study was multiplied by each inpatient admission from our data set. The estimation of annual hospitalisation costs per patient was carried out using a two-part model, with the first part estimating the probability of incurring any healthcare costs using a logistic regression model.
The second part estimated costs conditional on having incurred costs using a generalised linear model with a log link and a gamma distribution. Adjustments were made for age, sex, length of stay (LOS), comorbidities (Charlson Comorbidity Index (CCI)), type of stroke, rehabilitation, thrombolytic therapy with intravenous recombinant tissue plasminogen activator (rt-PA), type of hospital, health region and year of admission. Interaction terms between age and CCI were included in the model based on clinical evidence that most comorbid diseases become more common as people age (online supplemental table 3). These covariates were selected based on a review of the literature and clinically relevance. For the first modelling part, variables that were expected to impact on resource utilisation were included. The second part (cost estimation) included all variables used in the first part and in addition variables that were expected to affect costs (online supplemental table 4).
All-cause mortality
A Cox proportional hazards model was estimated initially, but showed violation of the proportional hazards assumption. Therefore, the Kaplan-Meier survival analysis and parametric survival analysis using the Gompertz distribution was employed.15 16 Survival time was measured in days from incident stroke until date of death or censoring date (latest recorded discharge date). Adjustment for covariates were similar to those used in the cost estimation, except for year of admission. A variable indicating recurrence of stroke was also added to the model. The interaction term between age group and CCI was included based on clinical evidence that most comorbid diseases become more common as people age (online supplemental tables 3 and 5).
All analyses were carried out using R software V.3.2 with the exception of the two-part models which were estimated using Stata V.14.
Patient and public involvement
Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Ethics approval
The ethics approval was not required for this study. Patient consent was not required because it is a retrospective study of an anonymised dataset.
Results
A total of 386 484 patients with stroke (first stroke) were identified from the database (table 1). Ischaemic stroke accounted for 50% (n=192 414) and haemorrhagic stroke accounted for 20% of all strokes; with the remaining 30% being recorded as unspecified stroke. Overall, 56% of the cohort were men. The mean age at incident stroke was 65 years, and was consistent across all stroke subtypes. More than 60% of patients with haemorrhagic stroke and ischaemic stroke had no comorbidities (CCI score 0) while 68% of patients with unspecified stroke had a CCI score of 1–2 which might imply that these patients could be more severely ill than patients with other subtypes.
Table 1Baseline characteristics of patients at incident stroke
| Haemorrhage (N=78 081) | Ischaemic (N=192 414) | Unspecified (N=115 989) | Total (N=386 484) | |
| Age (mean, SD) | 61 (15) | 66 (13) | 67 (14) | 65 (14) |
| 59 (14) | 64 (13) | 66 (14) | 63 (14) | |
| 64 (15) | 67 (14) | 69 (14) | 67 (14) | |
| Men (N, %) | 47 334 (61) | 103 261 (54) | 64 113 (55) | 214 708 (56) |
| Charlson Comorbidity Index (N, %) | ||||
| 47 528 (61) | 126 531 (66) | 32 920 (28) | 206 979 (54) | |
| 30 058 (38) | 64 024 (33) | 79 041 (68) | 173 123 (45) | |
| 495 (1) | 1859 (1) | 4028 (3) | 6382 (2) | |
N, number of patients.;
Resource utilisation
The mean LOS for incident stroke admission was greater among those with haemorrhagic stroke (9.9 days) compared with ischaemic stroke (6.1 days). Advanced-level hospitals recorded most of the incident strokes (50% of all haemorrhagic and 37% of all ischaemic strokes). The percentage of ischaemic patients receiving rt-PA was recorded at 7%. Only one-third of patients with stroke received rehabilitation during their incident stroke and ischaemic patients were more likely to receive rehabilitation compared with others. Computerised Tomographic (CT) scans, rt-PA prescription and rehabilitation services were mostly provided at advanced-level hospitals. Additionally, patients had four outpatient visits and two inpatient admissions on average. This was consistent across stroke subtypes. Average frequency of recurrent stroke, after excluding patients who died during their incident stroke, was approximately one (online supplemental table 6).
Cost estimation
Mean annual cost per patient was estimated to be 37 179 Baht (95% CI: 36 988 to 37 371). Haemorrhagic patients incurred higher costs compared to other subtypes (table 2).
Table 2Mean annual cost per patient by stroke subtypes
| Variable | All subtypes | Haemorrhagic | Ischaemic | Unspecified |
| mean (95% CI) | mean (95% CI) | mean (95% CI) | mean (95% CI) | |
| Stroke subtype | ||||
| 48 599 (48 099 to 49 099) | – | – | – | |
| 34 125 (33 879 to 34 371) | – | – | – | |
| 34 629 (34 284 to 34 974) | – | – | – | |
| Age group | ||||
| 34 752 (33 903 to 35 601) | 45 335 (44 187 to 46 482) | 31 833 (31 030 to 32 636) | 32 303 (31 451 to 33 155) | |
| 32 458 (31 916 to 33 001) | 42 340 (41 567 to 43 114) | 29 730 (29 201 to 30 259) | 30 170 (29 592 to 30 748) | |
| 34 375 (33 970 to 34 779) | 44 873 (44 228 to 45 517) | 31 508 (31 099 to 31 918) | 31 974 (31 501 to 32 447) | |
| 37 562 (37 208 to 37 917) | 49 017 (48 375 to 49 660) | 34 419 (34 052 to 34 785) | 34 927 (34 484 to 35 370) | |
| 40 111 (39 722 to 40 499) | 52 315 (51 602 to 53 028) | 36 734 (36 341 to 37 128) | 37 277 (36 802 to 37 752) | |
| 41 497 (41 008 to 41 986) | 54 113 (53 286 to 54 940) | 37 997 (37 518 to 38 475) | 38 558 (38 006 to 39 111) | |
| 41 187 (39 987 to 42 387) | 53 710 (52 057 to 55 362) | 37 714 (36 603 to 38 824) | 38 271 (37 112 to 39 430) | |
| Sex | ||||
| 36 959 (36 690 to 37 229) | 48 313 (47 749 to 48 877) | 33 924 (33 626 to 34 222) | 34 425 (34 040 to 34 810) | |
| 37 366 (37 116 to 37 616) | 48 845 (48 307 to 49 382) | 34 297 (34 008 to 34 587) | 34 804 (34 426 to 35 183) | |
| Charlson Comorbidity Index | ||||
| 30 086 (29 860 to 30 312) | 39 232 (38 806 to 39 659) | 27 548 (27 319 to 27 777) | 27 955 (27 597 to 28 314) | |
| 44 448 (44 127 to 44 768) | 57 923 (57 228 to 58 617) | 40 672 (40 300 to 41 044) | 41 273 (40 853 to 41 693) | |
| 70 242 (67 842 to 72 641) | 91 482 (88 226 to 94 738) | 64 236 (62 031 to 66 441) | 65 185 (62 975 to 67 396) | |
| Length of stay (LOS) | ||||
| 29 087 (28,837 to 29,336) | 38 525 (38 012 to 39 038) | 27 051 (26 794 to 27 309) | 27 451 (27 123 to 27 780) | |
| 29 782 (29 545 to 30 019) | 39 448 (38 961 to 39 935) | 27 699 (27 446 to 27 953) | 28 109 (27 780 to 28 437) | |
| 44 395 (43 916 to 44 875) | 58 793 (57 999 to 59 587) | 41 283 (40 777 to 41 789) | 41 893 (41 318 to 42 468) | |
| 93 221 (92 040 to 94 403) | 123 421 (121 689 to 125 153) | 86 663 (85 434 to 87 891) | 87 944 (86 574 to 89 313) | |
| Rehabilitation | ||||
| 38 490 (38 258 to 38 722) | 50 322 (49 794 to 50 850) | 35 335 (35 038 to 35 632) | 35 857 (35 497 to 36 217) | |
| 34 685 (34 381 to 34 988) | 45 347 (44 761 to 45 933) | 31 841 (31 551 to 32 131) | 32 312 (31 892 to 32 731) | |
| Thrombolysis | ||||
| 37 021 (36 828 to 37 213) | 48 375 (47 880 to 48 870) | 33 967 (33 718 to 34 217) | 34 469 (34 126 to 34 813) | |
| 41 986 (40 992 to 42 980) | 54 863* (53 429 to 56 297) | 38 523 (37 618 to 39 429) | 39 093 (38 098 to 40 087) |
*From 178 (0.05%) out of 386 484 patients.
95%CI, 95% confidence interval.
Key variables that significantly contributed to an increase in costs were found to be higher age, longer LOS, higher CCI score, receiving rt-PA at incident admission and being admitted to a hospital outside of the Bangkok area. Patients who received rehabilitation were estimated to incur lower costs than those who did not receive rehabilitation during their incident stay. Mean annual costs increased with increasing age from the age of 50 years and these figures showed a similar trend in all stroke subtypes. Patients who had a CCI score of ≥3 incurred costs twice as high as comparable patients who had 0 CCI score. Having longer LOS, especially >7 days, was associated with a statistically significant increase in costs compared with having shorter hospital stays of LOS <3 days (reference). Being admitted to non-MOPH facilities was associated with higher costs than being admitted to MOPH hospitals. Lastly, patients who received rt-PA at their incident admission incurred higher costs compared with patients who did not receive rt-PA.
Mean annual costs incurred by patients receiving rehabilitation at their incident admission were estimated to be lower (by 3806 Baht) compared with patients who did not receive rehabilitation. There were three health regions, where patients incurred higher costs compared with the Bangkok area (reference); however, mean costs were almost identical in all health regions (figure 1). Full results can be found in supplementary materials (online supplemental table 7).
Enlarge this image.Figure 1. The mean annual cost per patient classified by stroke subtypes and health regions. Health region 13: Bangkok
All-cause mortality
The Kaplan-Meier curves show the 4-year survival probability of all stroke subtypes (figure 2A, black line) was 66.5% (95% CI: 64.3% to 66.7%). There is a clear trend of decreasing survival probability during the 4 years following an incident stroke, with the ischaemic group having the highest probability of survival (70.5%; 95% CI: 70.2% to 70.7%) compared with other subtypes (figure 2A; unspecified: 60.6%; 95% CI: 60.2% to 61.0%; haemorrhagic: 64.9%; 95% CI: 64.4% to 65.5%). However, patients with haemorrhagic stroke had the lowest probability of survival at 1 year (76.5%; 95% CI: 76% to 77%) compared with other stroke subtypes (figure 2B).
Enlarge this image.Figure 2. Kaplan-Meier curves for all-cause mortality. (A) 4-year survival probability, (B) 1-year survival probability
After covariate adjustment (table 3, see online supplemental table 8 for full model results) the risk of mortality increased remarkably with age, and there was a noticeable upward trend in the risk of mortality especially in patients aged >70 years. This ranged from 2.5 to 15.6 times compared with those aged <40 years (reference), with patients with ischaemic stroke having a higher risk of mortality than patients with other stroke subtypes. The risk of mortality of patients who had a CCI score of 1–2 and a CCI score ≥3 were twice (95% CI: 1.9 to 2.3) and more than five times (95% CI: 4.2 to 7.9) the risk than in patients with no comorbidities. However, patients with stroke with a higher CCI score had a higher risk of mortality. There was also a higher risk of mortality for patients with longer LOS as compared with patients who had shorter hospital stays of <3 days. Patients with ischaemic stroke, whose LOS was ≥16 days, had 3.5 times the mortality risk (95% CI: 3.4 to 3.7) of patients who had LOS <3 days (reference). Further, patients who had a recurrent stroke had a 28% increase in the mortality risk compared with patients who had no recurrent stroke (95% CI: 1.26 to 1.31). Lastly, almost all geographical areas were associated with a higher risk of mortality compared with the Bangkok area.
Table 3Hazard ratios (HR) from the Gompertz model
| Covariates | Overall | Haemorrhage | Ischaemic | Unspecified |
| HR (95% CI) | HR (95% CI) | HR (95% CI) | HR (95% CI) | |
| Haemorrhage | Reference | |||
| Ischaemic | 0.76 (0.75 to 0.78) | |||
| Unspecified | 0.77 (0.76 to 0.79) | |||
| Women | Reference | |||
| Men | 1.07 (1.06 to 1.09) | 1.01 (0.97 to 1.04) | 1.06 (1.04 to 1.08) | 1.12 (1.10 to 1.15) |
| Age <40 | Reference | |||
| Age 40–49 | 1.17 (1.07 to 1.28) | 1.19 (1.04 to 1.37) | 1.25 (1.08 to 1.44) | 1.15 (0.94 to 1.41) |
| Age 50–59 | 1.57 (1.45 to 1.71) | 1.53 (1.35 to 1.74) | 1.77 (1.54 to 2.02) | 1.45 (1.21 to 1.74) |
| Age 60–69 | 2.47 (2.28 to 2.68) | 2.46 (2.18 to 2.79) | 2.76 (2.42 to 3.15) | 2.29 (1.92 to 2.73) |
| Age 70–79 | 4.58 (4.23 to 4.96) | 4.45 (3.93 to 5.03) | 5.11 (4.47 to 5.83) | 4.26 (3.58 to 5.07) |
| Age 80–89 | 8.48 (7.83 to 9.19) | 8.01 (7.07 to 9.08) | 9.62 (8.43 to 10.99) | 7.3 (6.12 to 8.71) |
| Age ≥90 | 15.59 (14.24 to 17.06) | 13.23 (11.26 to 15.55) | 18.00 (15.60 to 20.75) | 13.53 (11.04 to 16.59) |
| Charlson Comorbidity Index (CCI): score 0 | Reference | |||
| CCI: score 1–2 | 2.12 (1.92 to 2.34) | 1.71 (1.45 to 2.00) | 2.72 (2.29 to 3.22) | 1.85 (1.51 to 2.25) |
| CCI: score ≥3 | 5.77 (4.18 to 7.97) | 5.8 (2.58 to 13.02) | 4.90 (2.85 to 8.42) | 5.66 (3.50 to 9.16) |
| No rehabilitation | Reference | |||
| Received rehabilitation | 0.85 (0.84 to 0.86) | 0.75 (0.73 to 0.78) | 0.87 (0.85 to 0.88) | 0.92 (0.89 to 0.95) |
| No thrombolysis | Reference | |||
| Received thrombolysis | 0.93 (0.89 to 0.96) | 1.02 (0.74 to 1.40) | 0.86 (0.82 to 0.89) | 0.90 (0.73 to 1.10) |
| No recurrent stroke | Reference | |||
| Recurrent stroke | 1.28 (1.26 to 1.31) | 1.14 (1.09 to 1.20) | 1.36 (1.33 to 1.40) | 1.23 (1.19 to 1.28) |
| Length of stay (LOS) <3 days | Reference | |||
| LOS 3–7 days | 1.16 (1.14 to 1.18) | 0.63 (0.61 to 0.66) | 1.32 (1.29 to 1.35) | 1.21 (1.18 to 1.24) |
| LOS 7–15 days | 1.82 (1.78 to 1.85) | 0.76 (0.72 to 0.79) | 2.56 (2.49 to 2.63) | 1.81 (1.75 to 1.86) |
| LOS ≥16 days | 2.45 (2.39 to 2.51) | 1.13 (1.08 to 1.18) | 3.54 (3.41 to 3.66) | 2.40 (2.30 to 2.50) |
| Hospital type: primary and community hospital | Reference | |||
| Mid-level | 0.93 (0.91 to 0.96) | 1.05 (0.96 to 1.14) | 1.00 (0.96 to 1.04) | 0.87 (0.83 to 0.90) |
| Standard-level | 1.00 (0.98 to 1.02) | 1.10 (1.03 to 1.17) | 1.08 (1.05 to 1.12) | 0.94 (0.91 to 0.97) |
| Advanced-level | 0.91 (0.89 to 0.92) | 0.99 (0.94 to 1.06) | 1.01 (0.98 to 1.04) | 0.86 (0.83 to 0.88) |
| Non-Ministry of Public Health | 0.73 (0.70 to 0.75) | 0.83 (0.75 to 0.91) | 0.76 (0.71 to 0.81) | 0.66 (0.62 to 0.70) |
| Private hospitals/clinics | 0.99 (0.94 to 1.04) | 1.02 (0.91 to 1.16) | 1.05 (0.97 to 1.13) | 0.95 (0.88 to 1.03) |
Three additional measures were associated with a reduction in mortality risk. First, patients receiving rehabilitation during the incident episode showed a lower risk of mortality (0.85; 95% CI: 0.84 to 0.86) than patients who had no rehabilitation. Second, receiving rt-PA seemed to be associated with better health outcomes as it showed around 7% reduction in mortality (95% CI: 0.89 to 0.96). Third, among the types of hospitals, only patients admitted to non-MOPH hospitals showed a lower risk of mortality in all stroke subtypes when compared with those being admitted to primary/community hospitals (reference). When comparing between all subtypes, only patients with unspecified strokes had a reduced risk of mortality when being admitted, across hospital types, namely middle-level, standard-level, advanced-level and non-MOPH hospitals, with the only exception being private hospitals/clinics.
Discussion
This study is the first comprehensive analysis of recent Thai national stroke data to investigate costs and all-cause mortality of a nationally representative stroke cohort. Our results show that, presence of haemorrhagic stroke was associated with higher mean annual costs and 1-year risk of death compared with other stroke subtypes. Only one-third of patients with stroke received rehabilitation during their incident stroke and the percentage of thrombolysis was 7% for patients with ischaemic strokes. Possible explanations for a low rate of thrombolytic therapy10–12 could relate to (a) the onset of symptoms had been more than 4.5 hours7 17 which might be affected by health literacy of patients and families,18 or (b) patients may have had contraindications or poor prognosis, which could affect the rate of thrombolysis initiation. Additionally, costs of thrombolytic therapy may have played a role in the mean annual costs as patients who received thrombolytic therapy had higher mean annual costs18 19 compared with patients who did not receive thrombolytic therapy; however, it also played a vital role in improvement of mortality outcomes.
Our results support previous findings that although haemorrhagic is less common than ischaemic stroke, the cost that these patients incur tends to be higher.18 20 21 This could be because haemorrhagic strokes are associated with a poorer prognosis,22 23 higher risk of mortality,18 24 and requiring more resources, such as longer hospitalisations,18 to treat patients. Costs tended to be higher in older age groups in all stroke subtypes and costs incurred by patients with haemorrhagic stroke was highest in all age groups. We also found that patients with haemorrhagic stroke were younger than patients with ischaemic stroke. Furthermore, mean LOS in our study was shorter than other studies, ranging from 10 to 40 days21 25 26 but our finding is consistent with another Thai study.18 The shorter LOS is likely to be related to the diagnostic-related groups (DRGs) concepts to achieve cost containment, while the differences between other countries could be due to the variation of periods that counted after their hospitalisation.
We observed a low proportion of patients accessing rehabilitation services. This could be due to loss to follow-up while patients were being transferred to other healthcare settings, or lack of awareness of patients towards the importance of rehabilitation.5 Moreover, this finding also showed that patients with stroke did not receive inpatient rehabilitation properly in current practices but the new policy recommends inpatient rehabilitation services.27 These should focus further on cost-effectiveness and HRQoL such as the Barthel index scores, which has been suggested in the new rehabilitation guideline. However, patients receiving rehabilitation incurred lower costs than those who did not receive any rehabilitation. This could be partly explained by a less costly DRG value when discharged, with the reimbursement rate being lower than in the non-rehabilitation group. Another possible explanation is that there could be selection of faster recovering patients (with fewer comorbidities), who have the potential to gain more benefit from rehabilitation in real-life practices.27
Patients with ischaemic and unspecified stroke had a reduced mortality risk compared with patients with haemorrhagic stroke. Our findings related to an increased risk of mortality for older patients with stroke is consistent with a recent Thai study which indicates that stroke in the elderly is associated with higher mortality.28 Longer LOS was also associated with an increased mortality risk. These results may be explained by the fact that shorter LOS might be associated with lower risk of mortality or less severe strokes.22 29 Also, patients in regions other than Bangkok had a higher risk of death. This is comparable with previous Thai studies.28 As quality of care may have an effect on stroke survival, this inequality between health regions could be attributed to the differences in stroke management systems. Moreover, a scarcity of specialists in some health regions as well as the differences in advanced medical technologies could be attributed to differences in the quality of stroke care.28 29 Receiving rehabilitation was associated with a 15% decrease in the risk of death. Previous research reported that early rehabilitation is beneficial after stroke in the short-term and long-term.30–32
These results provide important insights into the different associations of our included covariates when mortality risk is modelled separately by stroke subtype and reveals findings that were masked when considering all stroke subtypes together. This information will be useful for policymakers for stroke management of specific subtypes in Thailand. Special attention for the service plan strategy should be given to the following activities: (1) follow-up on national KPI assessments in terms of health outcomes of stroke survivors to decrease costs and long-term risk of mortality, (2) improvement of rehabilitation post-hospital discharge as well as a daily functioning measurement (eg, recording of the Barthel Index scores in the national level database), (3) improvement of the health information system, linkage for interhospital transfer and continuum of care and (4) ensuring equitable access to care in all geographical areas.
Strengths and limitations
Our study provides results of mean annual cost and all-cause mortality of all stroke subtypes with the most up-to-date nationally representative stroke cohort. Although there are several published studies assessing the national stroke data, these mostly focused on costs or mortality only in specific subtypes and used out-of-date data.19 28 33 Our study has some limitations. The current data covered only UCS patients and did not include patients who seek treatment at non-contracted hospitals. Data on endovascular thrombectomy were also not available for the covered population. The analysis was based on hospitalised stroke events; however, we could have missed a small proportion of patients with less severe symptoms who were not admitted to hospital. Cause of death and clinical outcome measures, for example, functional scores, could not be obtained. We were therefore unable to evaluate functional disability or conduct competing event analysis. Mortality was analysed based on in-hospital mortality only, patients dying at home were not included. Finally, this study made an assumption on the history of previous strokes to determine incident stroke based on a 2-year lookback period, rather than clinical history of patients.
Conclusion
This study shows that crucial variables that were significantly associated with increasing costs or risk of mortality included being admitted to non-MOPH hospitals and being treated at health regions outside Bangkok. Importantly, rehabilitation might help save costs as well as contribute to a reduction in the risk of mortality. The measurement and recording of proven health outcomes measures of rehabilitation in the national level database, such as the Barthel scores, should be emphasised. The findings also revealed key differences between stroke subtypes which could help determine measurements for stroke management towards mitigation of costs and to ensure that the quality of stroke services is adequate to preserve or improve health outcomes of patients with stroke.
The authors would like to express deep gratitude and sincere appreciation to Mrs Netnaphis Suchonwanich and Mr Poonchana Wareechai (NHSO coordinator) for their valuable support.
Data availability statement
Data may be obtained from a third party and are not publicly available. We are unable to share the data used in this study since the data have been under a non-disclosure agreement with the National Health Security Office, Thailand.
Ethics statements
Patient consent for publication
Not applicable.
Ethics approval
The ethics approval was not required for this study. Patient consent was not required because it is a retrospective study of an anonymised dataset.
Contributors OW and CG conceptualised and designed the study. SK analysed the data and wrote the original draft manuscript. CG, OW and PL supervised, validated the analysis and reporting. All authors read, revised, edited and approved the final manuscript. SK is responsible for the overall content as guarantor.
Funding Health Policy and Systems Research program (HPSR Fellowship) under cooperation between National Health Security Office (NHSO), Bank for Agriculture and Agricultural Co-operatives (BAAC) and International Health Policy Program Foundation (IHPF). (Award/grant number: N/A).
Map disclaimer The inclusion of any map (including the depiction of any boundaries therein), or of any geographical or locational reference, does not imply the expression of any opinion whatsoever on the part of BMJ concerning the legal status of any country, territory, jurisdiction or area or of its authorities. Any such expression remains solely that of the relevant source and is not endorsed by BMJ. Maps are provided without any warranty of any kind, either express or implied.
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.
1 Strategy and Planning Division. Public health statistics 2018. Nonthaburi: Ministry of Public Health, 2019.
2 Johnson CO, Nguyen M, Roth GA, et al. Global, regional, and national burden of stroke, 1990-2016: a systematic analysis for the global burden of disease study 2016. The Lancet Neurology 2019; 18: 439–58. doi:10.1016/S1474-4422(19)30034-1
3 Institute for Health Metrics and Evaluation (IHME). GBD compare data visualization. Seattle, WA: IHME, University of Washington, 2020. Available: http://vizhub.healthdata.org/gbd-compare
4 National Health Security Office. Annual report 2019. Bangkok, 2020.
5 Khiaocharoen O, Pannarunothai S, Riewpaiboon W, et al. Rehabilitation service development for sub-acute and non-acute patients under the universal coverage scheme in Thailand. J Health Sci 2015; 24: 493–509.
6 Langhorne P, O’Donnell MJ, Chin SL, et al. Practice patterns and outcomes after stroke across countries at different economic levels (INTERSTROKE): an international observational study. Lancet 2018; 391: 2019–27. doi:10.1016/S0140-6736(18)30802-X
7 Suwanwela NC, Chutinet A, Kijpaisalratana N. Thrombolytic treatment in Thailand. J Stroke Med 2018; 1: 41–4. doi:10.1177/2516608518777934
8 Kijpaisalratana N, Chutinet A, Akarathanawat W, et al. Outcomes of thrombolytic therapy in acute ischemic stroke: Mothership, drip-and-ship, and ship-and-drip paradigms. BMC Neurol 2020; 20. doi:10.1186/s12883-020-1631-9
9 Dharmasaroja PA. Stroke network. J Thai Stroke Soc 2015; 14: 2.
10 Neurological information center. Report on service system development: stroke service plan 2018. Bangkok: Neurological Institute of Thailand, 2018.
11 Bureau of Health Administration. Service plan strategies 2018-2022. Nonthaburi: Office of the Permanent Secretary, Ministry of Public Health, Thailand, 2016.
12 Bureau of Inspection and Evaluation. Results of second round inspection and evaluation fiscal year 2018. Nonthaburi: Ministry of Public Health, 2018.
13 Winstein CJ, Stein J, Arena R, et al. Guidelines for adult stroke rehabilitation and recovery: A guideline for Healthcare professionals from the American heart Association/American stroke Association. Stroke 2016; 47: e98–169. doi:10.1161/STR.0000000000000098
14 Thai Casemix Centre, Thai Health Information Standards Development Center and Health Systems Research Institute. The unit cost per disease phase 1 (third year). Bangkok: Thai Casemix Centre, 2020.
15 Gray A, Clarke PM, Wolstenholme JL, et al. Applied methods of cost-effectiveness analysis in healthcare. Oxford University Press, 2010.
16 Collett D. Modelling survival data in medical research. CRC press, 2015. doi:10.1201/b18041
17 Rajborirug K, Tumviriyakul H, Suwanno J. Effects of stroke unit care in acute ischemic stroke patient ineligible for thrombolytic treatment. J Med Assoc Thai 2017; 100: 410–7.
18 Kongbunkiat K, Kasemsap N, Thepsuthammarat K, et al. National data on stroke outcomes in Thailand. J Clin Neurosci 2015; 22: 493–7. doi:10.1016/j.jocn.2014.08.031
19 Vongmongkol V, Tangcharoensthien V, Greetong T, et al. Trend in Recombinant tissue plasminogen activator (rtPA) use for ischemic stroke in Thailand: geographic inequality, cost of treatment and impact on 30-day case fatality rate. J Med Assoc Thai 2018; 101: 875–81.
20 Cha Y-J. The economic burden of stroke based on South Korea’s national health insurance claims database. Int J Health Policy Manag 2018; 7: 904–9. doi:10.15171/ijhpm.2018.42
21 Yoneda Y, Okuda S, Hamada R, et al. Hospital cost of ischemic stroke and intracerebral hemorrhage in Japanese stroke centers. Health Policy 2005; 73: 202–11. doi:10.1016/j.healthpol.2004.11.016
22 Andersen KK, Olsen TS, Dehlendorff C, et al. Hemorrhagic and ischemic strokes compared. Stroke 2009; 40: 2068–72. doi:10.1161/STROKEAHA.108.540112
23 Salvadori E, Papi G, Insalata G, et al. n.d. Comparison between ischemic and hemorrhagic strokes in functional outcome at discharge from an intensive rehabilitation hospital. Diagnostics; 11: 38. doi:10.3390/diagnostics11010038
24 Perna R, Temple J. Rehabilitation outcomes: ischemic versus hemorrhagic strokes. Behav Neurol 2015; 2015: 891651. doi:10.1155/2015/891651
25 Dodel RC, Haacke C, Zamzow K, et al. Resource utilization and costs of stroke unit care in Germany. Value Health 2004; 7: 144–52. doi:10.1111/j.1524-4733.2004.72314.x
26 Zhang H, Yin Y, Zhang C, et al. Costs of hospitalization for stroke from two urban health insurance claims data in Guangzhou city, Southern China. BMC Health Serv Res 2019; 19. doi:10.1186/s12913-019-4530-2
27 Intermediate Care Service Plan Board. Guideline for intermediate care. Thailand: Health Administration Division, Ministry of Public Health, 2019.
28 Butsing N, Mawn BE, Suwannapong N, et al. Long-term survival of stroke patients in Thailand. Southeast Asian J Trop Med Public Health 2018; 49: 322–30.
29 Suwanwela NC. Stroke epidemiology in Thailand. J Stroke 2014; 16: 1–7. doi:10.5853/jos.2014.16.1.1
30 Coleman ER, Moudgal R, Lang K, et al. Early rehabilitation after stroke: a narrative review. Curr Atheroscler Rep 2017; 19: 12.: 59. doi:10.1007/s11883-017-0686-6
31 Powers WJ, Rabinstein AA, Ackerson T, et al. Guidelines for the early management of patients with acute ischemic stroke: A guideline for Healthcare professionals from the American heart Association/American stroke Association. Stroke 2018; 49: e46–110. doi:10.1161/STR.0000000000000158
32 Wang W, Wei M, Cheng Y, et al. Safety and efficacy of early rehabilitation after stroke using mechanical Thrombectomy: A pilot randomized controlled trial. Front Neurol 2022; 13. doi:10.3389/fneur.2022.698439
33 Cumming K, Tiamkao S, Kongbunkiat K, et al. Impact of HIV on inpatient mortality and complications in stroke in Thailand: a national database study. Epidemiol Infect 2017; 145: 1285–91. doi:10.1017/S095026881600340X
© 2023 Author(s) (or their employer(s)) 2023. Re-use permitted under CC BY-NC. No commercial re-use. See rights and permissions. Published by BMJ. http://creativecommons.org/licenses/by-nc/4.0/ This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/ . Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.
Details
; Wu, Olivia 2 ; Langhorne, Peter 3 ; Geue, Claudia 2 1 Health Economics and Health Technology Assessment (HEHTA), School of Health and Wellbeing, University of Glasgow, Glasgow, UK; Health Intervention and Technology Assessment Program (HITAP), Ministry of Public Health, Nonthaburi, Thailand
2 Health Economics and Health Technology Assessment (HEHTA), School of Health and Wellbeing, University of Glasgow, Glasgow, UK
3 School of Cardiovascular and Metabolic Health, University of Glasgow, Glasgow, UK